we use pymongo 2.1.1, I've noticed such behavior in replica set (can
be easy reproduced):
scenario 1:
1) master host is shutted down by "shutdown -r now" command, or
hardware failure
2) slave becomes master
3) application fails saying it cant establish connection to the db
4) when the host with master's IP appears on the network and is
pingable (even without mongod on it), things become ok
scenario 2:
1) just mongod is shutted down on master host
2) slave becomes master
3) application handles failover and things are fine
is that normal behavior?